Address markj's comments.

The Transport group discussed this during a conference call and decided that this is good to go in.

Other than the two minor nits above this looks OK to me.

I like the use of DTrace as well. Please add the script as another commit.

Note that this has failed to build according to the build bot (see Harbormaster). Thoughts on that?

Other than the need for some clarifying parentheses I think this is OK.

Looks like Jenkins didn't like this little change either. You should look into the build failure. I also agree with the removal of _all() from the function name.

Other than the nits I pointed out already this looks like a good change.

Remove filtering based on system call type (suggested by rwatson)

New, BSD license, code does not need to be in cddl. Most of the new D scripts and programs go into either share/dtrace or the DTrace Toolkit port. Since this is a script I'd put it into share/dtrace if you want it in the src tree.

Actually it was in kern_fork() that I first looked, but, we do not have access to the syscall args structure at that point. I'll dig a bit more to see if I missed a pointer that would lead me there, or figure out if there is a reasonable way to synthesize one.

If you can wait a day or so I'd love to have Olivier, now added to the review, run one of his perf tests on this.

Address two of markj's suggestions.

Updated

This cannot be committed until the comments by julian (who you say proposed this) and others are addressed.

Address issue pointed out by mjg@

Is there any reason to remove the Illumos/Sparc code? Since it's under ifdef I'd leave that in for now.

There are a few things in here that seem to be both important and unresolved. I'll push np@ and gallatin@ to chime in because they have been the most frequent commenters. If this is not unstuck by Wednesday then let's get it in. But I hate having to "fix things in post."

One minor change to a name.

I agree it should be a separate option. Let's move this code forwards first. I've been talking to other developers about exposing hardware timestamps and that definitely needs to work in a different way.